Graphdb uložistě opakovaně vrací error 502

[blcham]

Na TEST deploymentu graphdb databaze opakovaně přestala fungovat a hlasi 502 Bad Gateway.

Popis problému

Ontographer hlási, že nemuze naloadovat slovníky.

Závažnost problému

• nemohu vůbec pracovat,

Reportovano od:

@bindeali

[blcham]

STATE OF THE SERVER

When looking at logs i saw many errors in al-db-server like the following:

al-db-server_1        | [WARN ] 2023-08-07 19:51:04,889 [repositories/assembly-line | o.e.r.h.s.r.t.ActiveTransactionRegistry] deregistered expired transaction d1a79952-eb8c-4fb1-af51-6ed94309c21d
al-db-server_1        | [WARN ] 2023-08-07 19:51:04,889 [repositories/assembly-line | o.e.r.h.s.r.t.ActiveTransactionRegistry] deregistered expired transaction 8be1860a-a480-4e0c-ad44-25eaf1534116
al-db-server_1        | [WARN ] 2023-08-07 19:51:04,889 [repositories/assembly-line | o.e.r.h.s.r.t.ActiveTransactionRegistry] deregistered expired transaction 1315a914-133b-4867-8022-7ce9e9f4e5dc
al-db-server_1        | [WARN ] 2023-08-07 19:51:04,889 [repositories/assembly-line | o.e.r.h.s.r.t.ActiveTransactionRegistry] deregistered expired transaction 920934fa-832c-45fb-804b-155d966f29f0
al-db-server_1        | [WARN ] 2023-08-07 19:51:04,889 [repositories/assembly-line | o.e.r.h.s.r.t.ActiveTransactionRegistry] deregistered expired transaction 1fa2e64d-678f-4f13-8142-cb0046f6d212

• docker stats:

• Memory configuration + amount of memory free:

---

POSSIBLE ISSUE

Although I did not see any "Out of memory" exception in the logs, I assume that it could be the case. The server has quite a small amount of memory, as can be seen in the picture above:

• 7.8 Gb in RAM
• 3.8 Gb in SWAP (done through a file)

---

RESOLUTION (EXPERIMENT)

Currently, the server has 7.8 Gb "fast RAM" + 3.8 Gb "slow RAM". I will increase "slow RAM"=SWAP to 16 GB. This should solve the issue, but it could slow down AL-tools that work with GraphDB -- so watch out for that! A better solution would be to increase "fast RAM".

@bindeali In case it happens again, please try to record what you were doing with the Ontographer so we know more information and we can try to simulate the issue.

[blcham]

I increased "slow RAM" as I suggested above:

I suggest closing the issue in a month if both:

• the issue does not happen again
• there is no significant slow-down of AL tools such as Ontographer (in that case, we should increase "fast RAM")